三维激光扫描点云直棱特征点填补方法

Automatic packing of straight edge feature in 3D laser scanning point cloud

在三维激光扫描中按照一定的角度分辨率对扫描物体进行离散化采样时,由于物体表面点云中棱角特征模糊,若直接用其构建三维表面模型,在棱角处会导致严重失真,进而影响模型的分析和应用.针对此问题提出了物体棱角自动化填补方法,其技术路线是:首先对点云建立空间索引、精简数据;其次根据扫描点云的法向量,对扫描点云进行非监督分类,从而将属于一个平面的点分为一类;最后利用平面相交的办法填补直线棱角上的特征点.通过实验证明,该方法自动化程度高、填补结果准确,能够使得建立的三维模型与真实物体更加接近,模型分析结果更加准确.

According to certain angle resolution, we conduct discrete sampling of scanning the object in the process of 3D laser scanning. Due to the fuzzy corner of object surface point cloud, it will lead to serious distortion at the edges and corners, which will affect the analysis and application of the model if we build 3D surface model directly. This paper put forward a method of auto- mated packing corner for this problem. Its technical route is as follows： firstly, we establish a spatial index and simplify data for point clouds. Secondly, we carry on the non-supervised classification which based on the normal vector of point cloud, then judge that belong to a planar points. Finally, we fill up feature points on the straight edge and corner by the plane intersection method. The experimental results show that this method is high automation degree and has accu- rate filling results, which makes the 3D model closer to the real world and the model analysis result is more accurate.